Energy efficiency has become one of the key challenges for a large class of electronic systems. Longer time between battery recharges is highly desirable for battery-powered devices such as mobile phones, digital cameras, Internet tablets, and electronic organizers. Energy efficiency is also important for electronic systems powered from the electric grid since it may reduce power consumption and the cooling requirements. Power savings are possible because electronic systems generally have an idle state, when, for example, the processor can run in a low-power state. Thus, the correct estimation of the workload model plays an essential role in the decision of which and when a power state transition should be performed by the electronic system. This paper introduces a multisize sliding window workload estimation technique for dynamic power management (DPM) in nonstationary environments. This technique reduces both the effects of identification delay and sampling error present in the previous fixed-size sliding window approach. The system is modeled by discrete-time Markov chains and the model offers a rigorous mathematical formulation of the problem and allows one to obtain an excellent trade-off between performance and power consumption.